Upload folder using huggingface_hub

README.md

@@ -153,6 +153,58 @@ result = classifier("Captain Torres must infiltrate enemy lines while battling h
 print(result)
 ```
 
+## Example Classifications
+
+Here are sample classifications showing the model's predictions with confidence scores:
+
+### NONE
+
+**Simple Example:**
+> *"Margaret runs the village bakery, making fresh bread every morning at 5 AM for the past thirty years."*
+
+**Prediction:** NONE ✅ (confidence: 0.997)
+
+**Nuanced Example:**
+> *"Dr. Harrison performs routine medical check-ups with methodical precision, maintaining professional distance while patients share their deepest fears about mortality."*
+
+**Prediction:** NONE ⚠️ (confidence: 0.581)
+
+### INTERNAL
+
+**Simple Example:**
+> *"Emma struggles with overwhelming anxiety after her father's harsh criticism, questioning her self-worth and abilities."*
+
+**Prediction:** INTERNAL ✅ (confidence: 0.983)
+
+**Nuanced Example:**
+> *"The renowned pianist Clara finds herself paralyzed by perfectionism, her childhood trauma surfacing as she prepares for the performance that could define her legacy."*
+
+**Prediction:** INTERNAL ✅ (confidence: 0.733)
+
+### EXTERNAL
+
+**Simple Example:**
+> *"Knight Roderick embarks on a dangerous quest to retrieve the stolen crown from the dragon's lair."*
+
+**Prediction:** EXTERNAL ✅ (confidence: 0.717)
+
+**Nuanced Example:**
+> *"Master thief Elias infiltrates the heavily guarded fortress, disabling security systems and evading patrol routes, each obstacle requiring new techniques and tools to reach the vault."*
+
+**Prediction:** EXTERNAL ✅ (confidence: 0.711)
+
+### BOTH
+
+**Simple Example:**
+> *"Sarah must rescue her kidnapped daughter from the terrorist compound while confronting her own paralyzing guilt about being an absent mother."*
+
+**Prediction:** BOTH ⚠️ (confidence: 0.578)
+
+**Nuanced Example:**
+> *"Archaeologist Sophia discovers an ancient artifact that could rewrite history, but must confront her own ethical boundaries and childhood abandonment issues as powerful forces try to silence her."*
+
+**Prediction:** BOTH ✅ (confidence: 0.926)
+
 ## Limitations
 
 - **Domain:** Optimized for character descriptions in narrative fiction

generate_readme_examples.py

@@ -0,0 +1,160 @@
+#!/usr/bin/env python3
+"""
+Generate 8 synthetic examples for README with model predictions
+2 examples per class: simple + nuanced
+"""
+
+import torch
+from transformers import DebertaV2Tokenizer, DebertaV2ForSequenceClassification
+
+def get_examples():
+    """Define 8 synthetic examples: 2 per class (simple + nuanced)"""
+    return [
+        # NONE - Simple
+        {
+            "text": "Margaret runs the village bakery, making fresh bread every morning at 5 AM for the past thirty years.",
+            "expected": "NONE",
+            "type": "simple"
+        },
+        # NONE - Nuanced  
+        {
+            "text": "Dr. Harrison performs routine medical check-ups with methodical precision, maintaining professional distance while patients share their deepest fears about mortality.",
+            "expected": "NONE", 
+            "type": "nuanced"
+        },
+        # INTERNAL - Simple
+        {
+            "text": "Emma struggles with overwhelming anxiety after her father's harsh criticism, questioning her self-worth and abilities.",
+            "expected": "INTERNAL",
+            "type": "simple"
+        },
+        # INTERNAL - Nuanced
+        {
+            "text": "The renowned pianist Clara finds herself paralyzed by perfectionism, her childhood trauma surfacing as she prepares for the performance that could define her legacy.",
+            "expected": "INTERNAL",
+            "type": "nuanced"
+        },
+        # EXTERNAL - Simple
+        {
+            "text": "Knight Roderick embarks on a dangerous quest to retrieve the stolen crown from the dragon's lair.",
+            "expected": "EXTERNAL", 
+            "type": "simple"
+        },
+        # EXTERNAL - Nuanced
+        {
+            "text": "Master thief Elias infiltrates the heavily guarded fortress, disabling security systems and evading patrol routes, each obstacle requiring new techniques and tools to reach the vault.",
+            "expected": "EXTERNAL",
+            "type": "nuanced"
+        },
+        # BOTH - Simple
+        {
+            "text": "Sarah must rescue her kidnapped daughter from the terrorist compound while confronting her own paralyzing guilt about being an absent mother.",
+            "expected": "BOTH",
+            "type": "simple"
+        },
+        # BOTH - Nuanced
+        {
+            "text": "Archaeologist Sophia discovers an ancient artifact that could rewrite history, but must confront her own ethical boundaries and childhood abandonment issues as powerful forces try to silence her.",
+            "expected": "BOTH",
+            "type": "nuanced"
+        }
+    ]
+
+def predict_examples():
+    """Run predictions on all examples"""
+    print("Loading model...")
+    tokenizer = DebertaV2Tokenizer.from_pretrained('.')
+    model = DebertaV2ForSequenceClassification.from_pretrained('.')
+    
+    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+    model.to(device)
+    model.eval()
+    
+    class_names = ['NONE', 'INTERNAL', 'EXTERNAL', 'BOTH']
+    examples = get_examples()
+    
+    results = []
+    
+    print(f"Running predictions on {len(examples)} examples...\n")
+    
+    for i, example in enumerate(examples, 1):
+        text = example['text']
+        expected = example['expected']
+        example_type = example['type']
+        
+        # Predict
+        inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True, max_length=512)
+        inputs = {k: v.to(device) for k, v in inputs.items()}
+        
+        with torch.no_grad():
+            outputs = model(**inputs)
+            probabilities = torch.softmax(outputs.logits, dim=-1)
+            predicted_idx = torch.argmax(probabilities, dim=-1).item()
+            confidence = probabilities[0][predicted_idx].item()
+        
+        predicted = class_names[predicted_idx]
+        is_correct = predicted == expected
+        
+        result = {
+            'text': text,
+            'expected': expected,
+            'predicted': predicted,
+            'confidence': confidence,
+            'correct': is_correct,
+            'type': example_type
+        }
+        
+        results.append(result)
+        
+        status = "✅" if is_correct else "❌"
+        print(f"{status} Example {i} ({expected} - {example_type})")
+        print(f"  Predicted: {predicted} (confidence: {confidence:.3f})")
+        print(f"  Text: {text[:80]}...")
+        print()
+    
+    return results
+
+def format_for_readme(results):
+    """Format results for README inclusion"""
+    
+    # Group by class
+    by_class = {}
+    for result in results:
+        expected = result['expected']
+        if expected not in by_class:
+            by_class[expected] = []
+        by_class[expected].append(result)
+    
+    readme_content = """
+## Example Classifications
+
+Here are sample classifications showing the model's predictions with confidence scores:
+
+"""
+    
+    for class_name in ['NONE', 'INTERNAL', 'EXTERNAL', 'BOTH']:
+        if class_name in by_class:
+            readme_content += f"### {class_name}\n\n"
+            
+            for result in by_class[class_name]:
+                confidence_icon = "✅" if result['confidence'] > 0.7 else "⚠️" if result['confidence'] > 0.5 else "❌"
+                
+                readme_content += f"**{result['type'].title()} Example:**\n"
+                readme_content += f"> *\"{result['text']}\"*\n\n"
+                readme_content += f"**Prediction:** {result['predicted']} {confidence_icon} (confidence: {result['confidence']:.3f})\n\n"
+    
+    return readme_content
+
+if __name__ == "__main__":
+    results = predict_examples()
+    readme_section = format_for_readme(results)
+    
+    print("README Section:")
+    print("=" * 50)
+    print(readme_section)
+    
+    # Save to file
+    with open('readme_examples_section.txt', 'w') as f:
+        f.write(readme_section)
+    
+    print("Saved README section to 'readme_examples_section.txt'")

readme_examples_section.txt

@@ -0,0 +1,53 @@
+
+## Example Classifications
+
+Here are sample classifications showing the model's predictions with confidence scores:
+
+### NONE
+
+**Simple Example:**
+> *"Margaret runs the village bakery, making fresh bread every morning at 5 AM for the past thirty years."*
+
+**Prediction:** NONE ✅ (confidence: 0.997)
+
+**Nuanced Example:**
+> *"Dr. Harrison performs routine medical check-ups with methodical precision, maintaining professional distance while patients share their deepest fears about mortality."*
+
+**Prediction:** NONE ⚠️ (confidence: 0.581)
+
+### INTERNAL
+
+**Simple Example:**
+> *"Emma struggles with overwhelming anxiety after her father's harsh criticism, questioning her self-worth and abilities."*
+
+**Prediction:** INTERNAL ✅ (confidence: 0.983)
+
+**Nuanced Example:**
+> *"The renowned pianist Clara finds herself paralyzed by perfectionism, her childhood trauma surfacing as she prepares for the performance that could define her legacy."*
+
+**Prediction:** INTERNAL ✅ (confidence: 0.733)
+
+### EXTERNAL
+
+**Simple Example:**
+> *"Knight Roderick embarks on a dangerous quest to retrieve the stolen crown from the dragon's lair."*
+
+**Prediction:** EXTERNAL ✅ (confidence: 0.717)
+
+**Nuanced Example:**
+> *"Master thief Elias infiltrates the heavily guarded fortress, disabling security systems and evading patrol routes, each obstacle requiring new techniques and tools to reach the vault."*
+
+**Prediction:** EXTERNAL ✅ (confidence: 0.711)
+
+### BOTH
+
+**Simple Example:**
+> *"Sarah must rescue her kidnapped daughter from the terrorist compound while confronting her own paralyzing guilt about being an absent mother."*
+
+**Prediction:** BOTH ⚠️ (confidence: 0.578)
+
+**Nuanced Example:**
+> *"Archaeologist Sophia discovers an ancient artifact that could rewrite history, but must confront her own ethical boundaries and childhood abandonment issues as powerful forces try to silence her."*
+
+**Prediction:** BOTH ✅ (confidence: 0.926)
+